Mypy is a static type checker for Python. It aims to combine the benefits of dynamic (or "duck") typing and static typing. Mypy can type-check your code and find potential errors before you run your program. For more information on getting started with Mypy, you can refer to the Mypy documentation.
Django-stubs is a collection of type stubs for Django, allowing you to use Mypy to check your Django code. It provides type hints for Django’s core features, making it easier to catch errors early and improving the overall development experience. We added Django-stubs to our project to enhance type checking for our Django codebase. By doing this, we can catch potential issues early in the development process, improving code quality and reducing bugs.
We have added a make target named make mypy which allows developers to run Mypy from within the bc_obps folder. This target simplifies the process of running Mypy, ensuring that our codebase adheres to the specified type annotations.
We cannot add types to Django-ninja ModelSchemas because they are generated at runtime and cannot be typed. For more information, refer to the discussion here.
We do not run static type checking on tests. Our tests are focused on verifying functionality, and we only want to check production code for type correctness.
In some places, we use # type: ignore[attr-defined] because of using Django-ninja ModelSchema. This directive tells Mypy to ignore attribute definition errors, which are common when dealing with dynamically generated models.
We use @typing.no_type_check on some functions to disable type checking for those specific functions. This is useful when type checking is not feasible or would require significant changes to the code that are not currently practical.
The Any type can be used when you need to disable type checking for a particular variable or expression. It effectively tells Mypy to ignore type checking for that value, allowing for maximum flexibility. However, overusing Any can negate the benefits of static type checking, so it should be used sparingly.
You can use Any by importing it from the typing module and then specifying it as the type for a variable or function parameter. For example:
from typing import Any
def process_data(data: Any) -> None:
# Function implementation
passThe # type: ignore comment can be used to suppress Mypy type checking on a specific line of code. This is useful when you know that a particular line will raise a type checking error that you want to ignore. Common scenarios include third-party libraries with incomplete type hints or code that is difficult to type correctly.
You can place # type: ignore at the end of a line to tell Mypy to ignore type errors for that line. For example:
result = some_function() # type: ignoreYou can also specify the exact error to ignore by adding it in square brackets, which can make your intent clearer and avoid ignoring other types of errors inadvertently:
result = some_function() # type: ignore[return-value]